# Opaque Rule Orderings ### Will Styler - LIGN 111 --- ### Today's Plan - Teaching Feedback - A note on Neutralization - A vacuous term - Transparent vs. Opaque Rule Orderings - Counter-feeding and Counter-Bleeding --- ## Teaching Feedback --- ### Thanks for all the feedback! - I'm always adjusting things like speed, timing, etc - Slides are posted on the Syllabus (every lecture day is a clickable link!) - We'll go over the group material a bit more - "The memes are a bit outdated, I guess." --- ### I'm on it!
--- ## Neutralization --- ### We've been dealing with a lot of problems where a contrast 'disappears' - Two words *are* distinctive underlyingly - ... but element of phonology makes that contrast go away --- ### Russian Devoicing
Nom. sg.
Dat. sg.
Nom. pl.
Gloss
[rok]
[roɡu]
[roɡa]
‘horn’
[rak]
[raku]
[raki]
‘crayfish’
[porok]
[poroɡu]
[poroɡi]
‘threshold’
[porok]
[poroku]
[poroki]
‘vice’
--- ## Neutralization When a phonemic contrast 'disappears' in certain environments due to phonological rules - "These words are underlyingly different, but the same on the surface" --- ### Neutralization is *really* common - You've been dealing with it already - ... and it's time you had the word for it --- ... speaking of terms... --- ## Vacuous Application When a rule applies to a given form without having any effect --- ### Obstruent voicing rule
[-voice,-son] -> [+voice] / __V
-
--- ### Obstruent voicing rule
[-son] -> [+voice] / __V
-
--- ### Rules can apply to forms while having no effect - [-son] -> [+voice] / __V will apply to /da/, but not result in any change - [-syll] -> [-high] / [-high]C0__ will apply to /e/, but nothing changes - The output can be the same as the input, and the rule can be fine --- ### Adding feature X to a segment with feature X does nothing - So, feel free to make your rules a bit more compact --- ### This is different from when the the environment isn't present - [-son] -> [+voice] / __V doesn't affect /na/ - The rule doesn't have the opportunity to apply --- # Now, for something completely different! --- ... I know what you're thinking ... - ### "Will, I want solving phonological problems to be more complex!" --- ## OK! --- ### Lomongo - **Gliding**: [-low] -> [-syll] / __V - **Yet [oina] is present in the language on the surface!** --- ## Opaque Interactions --- ### We've looked at Single-Rule Problems - There's just one alternation in the dataset, or several alternations stemming from one rule --- ### We've looked at Multi-Rule problems *without interactions* - Several rules, but they affect different segments - (... or the same segments in different contexts) - Here, **the ordering of the rules doesn't matter** --- ### Then, there are interacting rules - Multiple rules affecting the same segment(s) or region(s) of the word - One of the rules applies not to the UR, but to the output of another rule! - Here, **the ordering of the rules is crucial** --- ### We've looked at "feeding" rule interactions - Where one rule *provides a context* for another rule to apply --- ### Aspiration and Approx. Devoicing Derivation (Final)
Underlying Form
/plæt/
/splæt/
/pæt/
/læs/
Aspiration
| pʰlæt |
--
| pʰæt |
--
Approx. Devoicing
| pʰl̥æt |
--
--
--
Surface Form
[pʰl̥æt]
[splæt]
[pʰæt]
[læs]
--- ### We've looked at "bleeding" rule interactions - Where one rule *removes a context* for another rule to apply --- ### English Plural Derivation (Final)
Underlying Form
/kæt + z/
/dag + z/
/dɪʃ + z/
V Epenthesis
--
--
| dɪʃɪz |
Voicing Assim.
| kæts |
--
--
Surface Form
[kæts]
[dagz]
[dɪʃɪz]
--- ### Feeding and Bleeding interactions are *transparent* - Both rules are "visible" - The rules seem to "always apply" on the surface - You can look at the surface forms and see that these rules are not violated! --- ### There are two other kinds of rule orderings which are *opaque* --- ## Phonological Opacity When phonological processes interact in ways that may obscure the applicability or application of a given process --- ### Opaque Rule Orderings - Counterfeeding order - One rule "fails to feed" another - Counterbleeding order - One rule "fails to bleed" another --- ### Counterbleeding and Counterfeeding rule orderings are *opaque* - Sometimes, you'll see something 'illegal' happening on the surface - Your rule appears to fail to apply in some cases - Your rule appears to over-apply in some cases - The rule exists, but it can appear to be violated or over-applied in some situations - It is not "surface true" - Opaque orderings are trickier, theoretically and practically --- ### These names are not great - "Counter" here means the counterfactual inverse - "Rule A *would* feed rule B if ordered first, but instead, Rule B comes first. - "Rule C *would* bleed rule D if ordered first, but instead, Rule D comes first. --- ## Counterfeeding Ordering Rule A *fails to feed* rule B - Rule A would feed rule B, but it applies too late! --- ### Let's look at a counterfeeding ordering! --- ### Lomongo - (Remember Wednesday?) - **Gliding**: [-low] -> [-syll] / __V - **C Deletion**: [+voi,-son] -> ∅ / V__ --- ### Lomongo
Underlying Form
/o+bina/
/o+isa/
/ba+bina/
[-low] -> [-syll] / \_\_V
--
| wisa |
--
[+voi,-son] -> ∅ / V\_\_
| oina |
--
| baina |
Surface Form
[oina]
[wisa]
[baina]
- ### The Gliding rule is *violated on the surface* by [oina] --- ### A rule exists, but isn't always followed -
--- ### Counterfeeding ordering is opaque - Rules should be given a chance to apply to all words in the language - ... but it doesn't seem to apply to some forms - The correct ordering *fails to feed* that rule - The rule *seems* like it was violated - When it actually *failed to apply* to an earlier, intermediate form of the word --- ### The Counterfeeding Motto - "This rule seems like it should have applied here, but it didn't!" --- ## Counterbleeding --- ### Counterbleeding Ordering Rule A *fails to bleed* rule B - Rule A would bleed rule B, but it applies too late! --- ### Low German - **Spirantization**: [-son,+voice] -> [+cont] / V__ - **Devoicing**: [-son] -> [-voice] / __# --- ### Low German
Underlying Form
/ta:g/
/ta:g+ə/
/haʊz/
[-son,+voice] -> [+cont] / V\_\_
| ta:ɣ |
| ta:ɣə |
--
[-son] -> [-voice] / \_\_#
| ta:x |
--
| haʊs |
Surface Form
[ta:x]
[ta:ɣə]
[haʊs]
- ### Spirantization *shouldn't affect a voiceless sound*, but it seems like it did on the surface! --- ### Counterbleeding ordering is opaque - Rules should have the same effects throughout the language - ... but it seems to be applying where it shouldn't - The correct ordering *fails to bleed* that rule - The rule *seems* like it applied where it shouldn't - ... but it actually just applied earlier, to an intermediate form where it should have --- ### The Counterbleeding Motto - "This rule *shouldn't* apply here, but it seems to have!" --- ### Let's try and identify a rule ordering! --- ### English tapping and raising! - **Tapping**: [+ant, -cont, -lat] -> [-cons] / __[+syll, -stress] - ... and in some dialects... - **Raising**: /aɪ/ -> [ʌɪ] / __[-syll,-voice] - (Feature specification and diphthongs is tricky) --- ### English Tapping and Raising - Dialect 1
Underlying Form
/waɪt+əɹ/
/waɪd+əɹ/
[+ant, -cont, -lat] -> [-cons] / \_\_[+syll, -stress]
| waɪɾəɹ |
| waɪɾəɹ |
/aɪ/ -> [ʌɪ] / \_\_[-syll,-voice]
--
--
Surface Form
[waɪɾəɹ]
[waɪɾəɹ]
- ### What kind of rule ordering is this? Do we have a name for what happened at the end? --- ### English Tapping and Raising - Dialect 1 - Here, tapping **bleeds** raising in "Whiter" - Raising could apply, but tapping removes the environment (__[-syll, -voice]) it would need - The result is neutralization ("whiter" and "wider" become homophones) --- ### English Tapping and Raising - Dialect 2
Underlying Form
/waɪt+əɹ/
/waɪd+əɹ/
/aɪ/ -> [ʌɪ] / \_\_[-syll,-voice]
| wʌɪtəɹ |
--
[+ant, -cont, -lat] -> [-cons] / \_\_[+syll, -stress]
| wʌɪɾəɹ |
| waɪɾəɹ |
Surface Form
[wʌɪɾəɹ]
[waɪɾəɹ]
- ### What kind of rule ordering is this? --- ### English Tapping and Raising - Dialect 1 - Here, tapping **counterbleeds** raising in "Whiter" - Tapping *would* bleed raising, but it applies too late! - It looks like the rule *overapplied* - You shouldn't get [ʌɪ] before a voiced sound - There's no longer any neutralization! - *Dialects can differ in rule ordering!* --- ### Western Basque - **Mid Raising** - [-low] -> [+high] / __V - **Low Raising** - [+low] -> [-low] / __V --- ### Western Basque
Underlying Form
/alaba+a/
/seme+e/
[-low] -> [+high] / \_\_V
--
| semie |
[+low] -> [-low] / \_\_V
| alabea |
--
Surface Form
[alabea]
[semie]
- ### What kind of rule ordering is this? --- ### Western Basque - Here, mid raising **counterfeeds** low raising - Low raising *would* feed mid-raising, but it happens too late! - It looks like the rule was violated - You shouldn't get [e] before a vowel, but we did! --- ### Opacity is really tricky! - You get things on the surface that break your rules - ... and rules that seem to apply more often than they should - Your rules may be true, but not *surface true* --- ### Opacity is also theoretically interesting - Learning a rule that's regularly violated might be tougher - Handling opacity is a major area of differentiation among phonological theories - This is a bit beyond this class, but really interesting! --- ### The problem solving process doesn't change! - You're still generating rules - Then tweaking their orderings until things work - Then including derivation tables to show explain your analysis (always!) - **But you should think about opacity!** - Your rules might not be broken, just their orderings. - *That form might not be an exception at all.* --- ### Wrapping up - Neutralization is where phonology causes contrasts to disappear - Rule interactions can be transparent, or opaque - Counterfeeding orderings result in rules which should have applied, but didn't - Counterbleeding orderings result in rules which seem to apply where they shouldn't - Opacity is a royal pain, but really cool! --- ### Next time - Some counter(C0)eeding interactions! ---
Thank you!